Simulation of the Signals Transmission in Primary Motor Cortex Based on Time delay Neural Networks

• Main Authors: Shih-Cheng Chuang, 莊士正
• Other Authors: Rong-Chin Lo
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/7v4gwh

碩士 === 國立臺北科技大學 === 電腦與通訊研究所 === 96 === In the thesis, we build a time delay neural network (TDNN) with back– propagation learning rule to model the signals of neurons transduction in cerebral cortex. This model simulated the mechanism for modify movements by the thalamus control. And actually use the model to analyze the reliability of signal from the cortex of rat with probes. According to this model, the input layer is about action command from thalamus, the output layer is the action response reacts with commands, and hidden layer is interprocess. In experiments, we regard the signals as input commands, to simulate the action command from thalamus. We take the signals to train. According to the scheme which different commands cause different responses, we simulate the eight probes within multiple channels to capture the value of neurons in the neural model. Through Independent Component Analysis (ICA) and Subtractive Clustering methods, the proposed methods can classify the signals. When we classify the independent components of similar action commands, they have high correlation with each other. In non-similar is in the opposite. It demonstrated this model has a structure like the processing in cerebral cortex, and to illustrate by the signals can differentiate varieties of actions.